Intermediate steps in the degradation of a specific abnormal protein in Escherichia coli.

In order to learn more about the process of intracellular protein degradation, we have attempted to identify in Escherichia coli intermediate steps in the rapid degradation of the incomplete /3-galactosidase polypeptide that results from the nonsense mutation, X90. The disappearance of this polypeptide (M, = approximately 120,000) was followed with gel electrophoresis in sodium dodecyl sulfate, and the auto+ complementation assay was used to detect polypeptides containing the NH,-terminal end ((Y region) of figalactosidase. In addition to theX90 gene product (fragment A), extracts of strain X90 also contain a smaller polypeptide (fragment B) that had a M, if approximately 90,000 and contained the (Y region. Following removal of inducer and addition of chloramphenicol, total auto-a activity decreased with a half-life of about 30 min. Fragment A disappeared with a half-life of about 6 min. Concomittantly, fragment B increased 2-fold and then decayed with a half-life of about 30 min. The degradation of fragment B appeared to be the rate-limiting step for the disappearance of total auto+ activity from the cell. Fragment B was shown to be an intermediate produced from fragment A by an internal cleavage. Thus, the initial step in the catabolism of abnormal protein seems to be an endoproteolytic attack. Loss of auto-a activity, like degradation of other proteins, requires metabolic energy. When cells were treated with cyanide or azide, degradation of both fragments A and B was blocked. Energy was therefore required for at least two early, sequential steps in protein degradation. The disappearance of fragments A and B was also studied in degmutants which show a reduced ability to degrade nonsense fragments. In these mutants, the level of fragment A was greater, and the relative amount of B much lower than that in deg+ parents. Thus, in the degmutant the conversion of fragment A to fragment B was much slower than in the parent strain. During degradation or upon treatment with energy inhibitors, no degradative intermediates smaller than B were


Intermediate
Steps in the Degradation of a Specific Abnormal Protein in Escherichia coZi* In order to learn more about the process of intracellular protein degradation, we have attempted to identify in Escherichia coli intermediate steps in the rapid degradation of the incomplete /3-galactosidase polypeptide that results from the nonsense mutation, X90. The disappearance of this polypeptide (M, = approximately 120,000) was followed with gel electrophoresis in sodium dodecyl sulfate, and the auto+ complementation assay was used to detect polypeptides containing the NH,-terminal end ((Y region) of figalactosidase. In addition to theX90 gene product (fragment A), extracts of strain X90 also contain a smaller polypeptide (fragment B) that had a M, if approximately 90,000 and contained the (Y region. Following removal of inducer and addition of chloramphenicol, total auto-a activity decreased with a half-life of about 30 min. Fragment A disappeared with a half-life of about 6 min. Concomittantly, fragment B increased 2-fold and then decayed with a half-life of about 30 min. The degradation of fragment B appeared to be the rate-limiting step for the disappearance of total auto+ activity from the cell. Fragment B was shown to be an intermediate produced from fragment A by an internal cleavage. Thus, the initial step in the catabolism of abnormal protein seems to be an endoproteolytic attack. Loss of auto-a activity, like degradation of other proteins, requires metabolic energy. When cells were treated with cyanide or azide, degradation of both fragments A and B was blocked. Energy was therefore required for at least two early, sequential steps in protein degradation. The disappearance of fragments A and B was also studied in degmutants which show a reduced ability to degrade nonsense fragments. In these mutants, the level of fragment A was greater, and the relative amount of B much lower than that in deg+ parents. Thus, in the deg-mutant the conversion of fragment A to fragment B was much slower than in the parent strain.
During degradation or upon treatment with energy inhibitors, no degradative intermediates smaller than B were found by these techniques. Several polypeptides containing auto-a activity were seen on gels, but they appear to be artifacts produced by boiling. When extracts of X90 were treated with mercaptoethanol and sodium dodecyl sulfate and heated to 100" for several minutes, partial cleavage of the polypeptide chain occurred.
In bacterial and animal cells, proteins with highly abnormal structures are degraded much more rapidly than are normal proteins (1, 2). This process serves to rid cells of denatured proteins, mutant polypeptides, proteins resulting from errors in synthesis, or prematurely terminated polypeptides resulting from umber or ochre mutations or from incorporation of puromycin (l-6). For example, in wild type Escherichia coli K12, P-galactosidase is a stable protein, which undergoes no detectable degradation (7, 8). However, a nonsense mutation (X90) in the z-gene leads to production of an incomplete polypeptide which differs in size only slightly from the wild type enzyme, but which is degraded by cells with a half-life of about 7 min (4). The mechanisms for recognition and selective hydrolysis of such abnormal proteins, the nature of the proteolytic process, and the identity of the responsible proteases are not known.
The present studies were undertaken to determine whether the initial cleavage reactions are exoproteolytic or endoproteolytic and to define specific steps in the degradative process. We have utilized mutant strains isolated by Bukhari and Zipser (5) that have a reduced ability to degrade nonsense fragments of @galactosidase.
Unfortunately, the nature of the defect in these strains has not yet been determined. One interesting, unexplained feature of the degradation of both normal and abnormal proteins is that this process can be blocked with inhibitors of energy metabolism (2,9,10). At first glance, this apparent energy requirement is somewhat surprising since proteolysis must be an exergonic process and since no known protease requires high energy cofactors. These experiments therefore attempted to determine whether high energy compounds are required only for an initial step in protein degradation or perhaps only for the conversion of peptides to amino acids.
In order to investigate these questions, we have attempted to identify intermediates in the intracellular degradation of a single protein, the nonsense fragment of p-galactosidase produced in E. coli X90. The characterization of such intermediates should be very useful in defining steps in the degradative

Intermediate
Steps in Degradation of an Abnormal Protein 8351 process. Such intermediates, however, might be difficult to detect since they lack enzymatic activity and may be degraded quite rapidly. Since no cell-free system has yet been described which duplicates protein degradation in vivo, these studies were performed with intact cells.
We chose to examine the degradation of a nonsense fragment of P-galactosidase because such inactive polypeptides as well as intermediates in their degradation may be measured by the auto-a complementation assay (11,12). Morrison and Zipser (11) first showed that when nonsense fragments are autoclaved, an NH,-terminal polypeptide ("auto-a"; M, = 7,400) is released. When auto-a is mixed with an "acceptor" polypeptide containing a deletion within the (Y region of pgalactosidase, the two inactive polypeptides combine to yield active enzyme (11,12). This technique permits a quantitative assay for nonsense fragments of p-galactosidase and for any of the degradative products that contain an intact NH,-terminal region.
Using this assay, Zipser and co-workers (13,14) found that strain X90 contained not one, but two, polypeptides containing the NH,-terminus of /3-galactosidase. The larger polypeptide had the expected molecular weight for the mutant gene product (M, = about 120,000'). They suggested several possible explanations to account for the presence of the smaller polypeptide (M, = 90,000). The present studies demonstrate that this smaller polypeptide is actually an intermediate in protein degradation, and attempt to define the early steps in this process.

MATERIALS AND METHODS
Strains and Media-Strains X90 (B1) and XA21 (F'lac i-zA2I/i-cA211 were kindly provided by Dr. I. Zabin, University of California, Los Angeles. Isogenic strains 8057 (F' pro Zac z X90/A pro Zoc deg T+) and 8058 (F' pro Zac z X90/A pro lac deg T-) were a gift of Dr. Ahmed Bukhari.
The minimal medium (15) was supplemented with vitamin Bl (10 pg/ml) and 0.5% glycerol or succinate was added as a carbon source. Preparation of Acceptor for Complementation Assays-Escherichia coZi strain XA21 containing a deletion within the 01 region of the pgalactosidase gene was grown in LB broth at 37" with aeration and harvested in log phase. All further steps were performed at O-4". The cell pellet (about 35 g wet weight) was suspended in Buffer I (20 rnM Tris.HCl, 5 rnM EDTA, 10 rnM NaCl, 1% mercaptoethanol, pH 7.2) and washed once by centrifugation for 15 min at 25,000 x g. The cells were suspended by addition of 80 ml of Buffer I and disrupted by repeated sonication (total time, 2 min) in a Branson sonifier cell disrupter.
After centrifugation for 20 min at 20,000 x g, the supernatant containing auto-a acceptor was decanted. The pellet was suspended by addition of 20 ml of Buffer I, sonicated as before, and centrifuged, and the supernatant was combined with the initial supernatant.
After centrifugation for 90 min at 100,000 x g to remove particulate matter, the supernatant (about 50 ml) was removed, and sucrose was added to it to give a final concentration of 10% (w/v). This solution containing acceptor activity was stored in 5-ml aliquots at -30". Before freezing, acceptor activity was measured with the complementation assay using a standard preparation of auto-a (see below).
Degradation Experiments -Cells were grown at 37" with aeration for at least three generations in the presence of 5 x 10Y5 M isopropylthio-/3-n-galactopyranoside.
During log phase, cells were collected by pouring over crushed ice and centrifuging for 12 min at 20,000 x g in the cold (4"). The cells were resuspended in 20 ml of ice cold medium lacking IPTGZ and centrifuged again at the same i The molecular weight of the larger polypeptide is indistinguishable from that of P-galactosidase on sodium dodecyl sulfate gels. Recent evidence from amino acid sequencing indicates a M, = 120,000 for /3-galactosidase (24). * The abbreviations used are: IPTG, isopropylthio-P-n-galactoside; SDS, sodium dodecyl sulfate; ONPG, o-nitrophenyl-P-n-galactopy-speed. The pellets were suspended in 5 ml of fresh cold medium and diluted (at time zero) into 20 to 30 ml of medium at 37", and degradation of auto-a activity at 37" was followed. At various times, samples (1 to 4 ml) were removed, pipetted onto ice, and centrifuged 15 min at 28,000 x g at 4". The supernatants were discarded and the cell pellets were frozen and stored at -30".
These samples were then either assayed directly for auto-o activity or subjected to gel electrophoresis.
Auto-a Assay-Auto-a assays were performed by a modification of the method of Morrison and Zipser (11). Frozen cell pellets were suspended in 0.5 ml of ice-cold Buffer II (20 rnM Tris.HCl (pH 7.21, 0.1 M NaCl, 1 mM MgCl,, 1% mercaptoethanol, 0.5 mM EDTA) in covered tubes and autoclaved for 40 min at 250" to produce the auto-LY polypeptide.
After centrifugation to remove the precipitates, supernatants were removed and either assayed directly or stored at -30". This material served as auto-a donor in the complementation assay.
To assay auto-o activity, 100 ~1 of auto-a donor and 50 ~1 of auto-Q acceptor were mixed and incubated at 28" for 90 min to permit the complementation reaction to occur. (The EDTA in the acceptor preparations aids in its stability but interferes with the assay. Therefore, prior to complementation, MgCl, was added to the acceptor solution to give a final concentration of 10 mM.) To assay the complemented enzyme, 0.5 ml of a stock solution (3 mM) of onitrophenyl-P-n-galactopyranoside was added. The ONPG was dissolved in Buffer II containing 10 mM sodium azide to prevent microbial growth.
The enzyme was incubated at 28", and when a yellow color had appeared, the reaction was stopped by addition of 0.5 ml of 1 M Na,C03.
Absorbance was read at 420 nm. One unit of enzyme activity equaled that amount which hydrolyzed 1 nmol of ONPG in 1 min at 28". Auto-a activity in a sample was defined as the activity of the complemented enzyme formed from that sample. The amount of ONPG hydrolyzed in this assay was proportional to the amount of auto-a containing extract added and to the time of incubation (data not shown).

SDS Gel Electrophoresis
and Determination ofAuto-a from Gels-Slab gels were prepared according to the methods of Laemmli (16). To prepare samples for gels, frozen cell pellets were dissolved in 0.25 ml of buffer III (0.065 M Tris.HCl (pH 6.8), 2% sodium dodecyl sulfate (Matheson, Coleman, and Bell, Technical), 2% mercaptoethanol, 10% (v/v)glycerol, 0.001% bromphenol blue) and heated at 100 for 5 min (unless otherwise stated). Samples of 25 ~1 were layered onto each slot of the slab gel along with 10 ~1 of crude paramyosin (a gift of Dr. John Weisel, Brandeis University, Waltham, Mass.) which had been dansylated with dansyl chloride (17) and which served as an internal molecular weight marker. Electrophoresis was carried out at 125 V (constant voltage) for 3l/2 h, during which the dye migrated approximately 100 mm.
Auto-a in the gels was assayed by a modification of the method of Morrison et al. (14). After electrophoresis, the gel was soaked in 500 ml of 15% ice cold trichloroacetic acid for 24 h with changes of trichloroacetic acid after 2, 5, and 10 h. This treatment fixed the proteins and extracted the SDS. The gel was then neutralized by soaking 15 min with 500 ml of cold buffer (containing 0.04 M Tris base, 0.02 M sodium acetate, 0.002 M sodium EDTA, and brought to pH 10 with NaOH) while shaking gently.
The gel was then soaked for 25 min in 250 ml cold Buffer IV (0.4 M Tris base, 0.2 M sodium acetate, 0.02 M sodium EDTA, brought to pH 7.6 with glacial acetic acid), for 10 min in 250 ml of fresh cold Buffer IV, and for 5 min in 250 ml of fresh cold Buffer IV which had been diluted I-fold with distilled water. After neutralization, the slab gel was examined under black light illumination which caused the dansylated markers to fluoresce.
To separate the electrophoretograms of the various samples, the slab gel was then cut parallel to the direction of migration along the nonfluorescent spacer'regions. Using a block of razor blades spaced 1 mm apart, each of these gels was then sliced transversely into l-mm pieces. Each piece was placed in a small test tube for the auto-a assay.

RESULTS
Initial experiments confirmed the rapid degradation of autoa-containing polypeptides in Escherichia coli strain X90, and demonstrated the existence in this strain of two protein fragments of p-galactosidase.
Cells were grown on glycerol minimal medium containing IPTG, washed to remove inducer,  Fig. 3A were removed at t = 0 (0-O) and t = 150 min CO---01, electrophoresed, and the distribution of auto-a activity on SDS acrylamide gels was measured. process, as has also been observed by Zipser and co-workers.3 By analyzing the levels of fragments A and B following energy inhibition, it was possible to determine whether metabolic energy was required for the initial conversion of A to B, for a later step, or throughout the degradative pathway.
The effects of cyanide on the levels of fragments A and B were analyzed by electrophoresis (Fig. 5) using the same cells studied in Fig. 4. In the absence of cyanide, fragment A decayed with a half-life of about 17 min or less (Fig. 5) Azide completely blocked degradation of total auto-a activity in the cells (Fig. 4), inhibited the loss of the fragments A and B, and did not cause the appearance of any new peaks (Fig.  6) example, clumping of cells occurred upon treatment with azide, but not with cyanide; also during azide treatment the abnormal polypeptides may have associated into dense intracellular aggregates (10, 18). This latter possibility is now being investigated.
Studies of Deg-Mutants-The deg-mutants (5) degrade nonsense fragments of P-galactosidase at reduced rates. To test where in the degradative process the mutations might be acting, we have analyzed the degradation of polypeptide fragments A and B in such mutants. Induced cells of deg+ and deg-carrying the X90 mutation were washed and suspended in inducer-free medium with chloramphehicol and degradation was followed (Fig. 7). The half-lives were 280 min for deg-and 71 min for deg+. Cell samples from this experiment were analyzed on gels in order to follow the fates of individual fragments. At time zero (Fig. 8), the deg+ strain contained fragments A and B in approximately a 1:l ratio in accord with earlier experiments (Fig. 2). In the deg-mutants, the absolute level of fragment A was at least 6-fold greater than in the parent strain (Fig. 8). This result suggests a reduced rate of degradation of fragment A. Furthermore, the deg-mutant had 86% of its auto-a activity in fragment A and only 14% in fragment B. No other peaks were detectable. This low proportion of auto-a activity in fragment B in degindicated that either the rate of degradation of B was greater in deg-than deg+ or that conversion of A to B was reduced in the mutant. Samples taken at various times resolved this question (Fig. 9). In the deg+ parent, fragment A decreased with a half-life of 8 min while B at first rose by 30% of its original value (due to continuous conversion of A to B) and then fell. In the mutant, however, the rate of decay of fragment A was greatly reduced. From these data, it was not possible to determine whether the degradation of fragment B was also inhibited even though the level of B decreased more slowly in deg-. Since A remained higher than B during the  Fig. 7 were removed at t = 0 and analyzed for auto-a activity following SDS gel electrophoresis. No other peaks of activity were found on these gels. experiment, synthesis of B from A may continue to occur, and the level of B reflects both its synthesis and degradation in the deg-.
More precise data during the later times are needed to determine if the leveling off of B represents continued synthesis of B from A or a slower rate of degradation of fragment B in the deg-. Furthermore, in the deg-strain, the loss of auto+ activity was much more rapid when measured from gels than when cells were assayed without electrophoresis (Fig.  7). After 90 min of degradation, 20% of auto-a activity was lost from the cells (Fig. 7), but a loss of 70% of activity in fragments A + B was found by gel analysis (Fig.  9) Steps in Degradation of an Abnormal Protein carboxyl fragment resulting from the cleavage. Since this molecule should not be detectable by the auto+ assay, we have attempted to find it using antibodies against P-galactosidase." These antibody experiments showed similar proportions of peak A and B and similar kinetics of decay as were obtained by assaying o-complementing activity. Although they confirmed the precursor-product relationship between A and B, they have as yet failed to detect a carboxyl fragment. (In the absence of such data, we cannot completely eliminate the possibility that an exoprotease removes amino acids very rapidly from the carboxyl end of fragment A, until fragment B remains.) cleaved in many places, and still retain its tertiary structure and enzymatic activity (21). However, there is no evidence that such strong binding may occur in the presence of SDS.
It has generally been assumed that an initial proteolytic cleavage is the rate-limiting step in the degradation of a protein to amino acids. However, the half-life of fragment B (about 30 min) was at least 5 times longer than that of A (Fig. 2B). Thus, decay of fragment B seems to be the ratelimiting step for disappearance of auto-a containing polypeptides from the cell. Several other nonsense mutants have been found to contain two polypeptide fragments of P-galactosidase, one corresponding in size to the presumed gene product and the other smaller (13,14). Fragment B should correspond to the map position 11 of Lin and Zabin (3) which corresponds to a M, = approximately 90,000. For those mutant strains in which the gene product should be larger than fragment B, the size of the smaller polypeptide resembles that of fragment B (13, 14), and the half-lives of auto-a activity in these cells approximate that of fragment B. For those mutants in which the nonsense fragments should be smaller than B, the halflives are much shorter with one exception (3). These observations suggest that in strains with nonsense fragments greater than M, = 90,000, an initial endoproteolytic cleavage always occurs at the same site to yield fragment B, plus a peptide whose size varies with the mutant. Fragment B is relatively stable compared to smaller auto-a containing polypeptides, whose half-lives may be as short as a few minutes (3). Thus the half-life of B should be rate-limiting for decay of auto-a activity in all strains with large nonsense fragments (map position 11 or greater).
The term "half-life" has been used a bit loosely in this paper and by others (3, 5) to describe the decay of a heterogeneous population of polypeptides all containing auto-a activity (e.g. Fig. 1) even though this process is in fact not strictly exponential. Lin and Zabin (3) first noted that the rate of decay of auto-a activity in strains carrying X90 or other nonsense mutations was not exponential, but decreased after a period of time which varied in different mutants. This finding has been confirmed in the present studies (Fig. 3A) which further show that this decrease in auto-a decay is not due to the accumulation of new stable polypeptides containing the auto-a region (Fig. 3B). Residual activity was found principally in fragments A and B. The degradation of fragments A (Figs. 5 and 9A) and B (data not shown) was also initially rapid but decreased markedly when these polypeptides were present in very low amounts.
It is interesting that these biphasic kinetics were obtained only in deg+ cells. In deg-strains, the initial rapid decay process appears to be lacking (Fig. 7). Possibly, these results suggest the existence of two proteolytic systems capable of attacking nonsense fragments, one responsible for the rapid degradation of auto-a polypeptides, that is defective in degcells, and one for the later, slower degradative process apparent in deg+ cells only with low levels of substrate. This laboratory has made extensive efforts to find differences between the proteolytic activity of extracts of deg+ and degstrains without success" (22).
These studies have failed to demonstrate any other smaller degradative intermediates in X90. Goldschmidt reported a M, = 60,000 polypeptide containing auto-a activity on SDS gels that represented 6% of the total auto-a activity in strain X90 (13). We have also found this peak and several additional smaller ones (Fig. 3B). However, the amount of auto-a activity in polypeptides with M, < 90,000 depended on the method of treatment of the samples. The more gentle the treatment prior to electrophoresis, the less activity was found in these low M, peaks (Table I). Thus, these polypeptides appeared to result from cleavage of both fragments A and B induced by boiling. To our knowledge, there are no previous reports of hydrolysis of proteins upon boiling in SDS (20). However, the formation of the auto-a polypeptide is by itself evidence that high temperatures can cause proteolytic cleavages at specific sites. In addition, boiling of commercially obtained pure pgalactosidase causes a decrease in the molecular weight of the polypeptide.' It is also possible that the small fragments (Table I) represent degradative intermediates which are tightly, but not covalently, bound to the complementary polypeptide, so that SDS treatment does not dissociate them without heating. It is known that /3-galactosidase can be The degradation of the auto-o containing polypeptides, like other cell proteins (9, lo), appears to require metabolic energy (Fig. 4). This process was blocked almost completely by both cyanide and azide which inhibit respiration.
In addition, cyanide (Fig. 5) and azide (data not shown) reduced both the conversion of A to B and the subsequent hydrolysis of fragment B. Therefore, energy is required for at least two and possibly more steps in the degradative pathway. Related studies in this laboratory indicate that energy is also required for breakdown of much smaller fragments of P-galactosidase (M, < 8,000) (22). It has been suggested that metabolic energy is required for compartmentalization of proteins in some degradative structures (e.g. periplasm or lysosome) or to activate the protease or for chemical modification of the proteins to make them more susceptible for degradation (2). In either case, the finding that energy is required for multiple steps in degradation would suggest that repeated modification of the substrate is necessary or that the protein must undergo repeated uptake into such a degradative compartment.